基于ARIMA-LSTM模型的综合能源系统负荷与风光资源预测
王 鑫1,李安桂1,李 扬1,卜令晨1,彭怀午2,牛东圣2,许晨琛2,韩 欧1

(1.西安建筑科技大学 建筑设备科学与工程学院，陕西 西安，710055; 2.太阳能利用工程技术研究所，中国电建集团西北勘测设计研究院有限公司，陕西 西安，710065)

综合能源系统; 负荷预测; 构建模型; 误差分析

Multivariate load prediction and wind-solar resource characteristic quantity prediction of integrated energy system based on ARIMA-LSTM model
WANG Xin1,LI Angui1,LI Yang1,BU Lingchen1,PENG Huaiwu2,NIU Dongsheng2,XU Chenchen2,HAN Ou1

(1.School of Building Services Science and Engineering, Xi'an Univ. of Arch. & Tech., Xi'an 710055, China; 2.Institute of Solar Engineering Technology, Northwest Engineering Corporation Limited, Power China, Xi'an 710065, China)

integrated energy system; load forecasting; construction model; error analysis

DOI: 10.15986/j.1006-7930.2022.05.015

备注

在能源互联网快速发展的背景下，研究分析了综合能源系统的多元负荷预测模型及理论方法.针对传统ARIMA(Autoregressive Moving Average Model，ARMA)模型仅能处理线性关系的问题，将ARIMA模型与LSTM(Long-Short Term Memory，LSTM)网络模型结合，提出并建立了ARIMA-LSTM模型.该模型不仅兼容冷、热、气、电等多元负荷的预测，并且可以用于风速、辐射照度等数据的预测，有较好的适应性和预测精度.
Under the background of the rapid development of energy Internet, this paper studies and analyzes the multivariate load prediction and theoretical method of integrated energy system. Aiming at the problem that the traditional ARIMA model can only deal with the linear relationship, the ARIMA-LSTM model is proposed and established by combining ARIMA model with LSTM( Long-Short Term Memory, LSTM )network model. The model is not only compatible with the prediction of multiple loads such as cold, heat, gas and electricity, but also can be used for the prediction of wind speed, radiation illumination and other data, and has good adaptability and prediction accuracy.